Predictive Analytics – QTA 405
End Term Project
RBA 2024-26
Marks :60
The dataset provided contains time series data representing a particular variable over a period.
Tasks:
Data Importing and Preprocessing:
Load the time series dataset into Python.
Check the structure and summary statistics of the dataset.
Handle any missing values or inconsistencies if present.
Time Series Visualization:
Create a time series plot to visualize the trend and seasonality in the data.
Plot other relevant graphs such as a histogram to understand the distribution of the data.
Moving Average:
Implement the moving average method to smooth the time series data.
Choose an appropriate window size for the moving average.
Plot the original time series data along with the smoothed data using moving average.
Exponential Smoothing:
Apply Simple/Double and Triple exponential smoothing to the time series data using different
smoothing parameters (e.g., alpha).
Compare the results of exponential smoothing with the original data and moving average.
Plot the original time series data along with the smoothed data using exponential smoothing.
ARIMA Modelling:
Conduct a time series decomposition to identify trend, seasonality, and residual components.
Determine the appropriate parameters (p, d, q) for the ARIMA model using autocorrelation
and partial autocorrelation plots.
Fit an ARIMA model to the time series data.
Validate the ARIMA model using diagnostic plots and statistical tests.
Forecast future values of the time series using the ARIMA model.
Prediction Task:
Select the best-performing model (moving average, exponential smoothing, or ARIMA) based
on forecasting accuracy.
Make predictions for future values of the time series using the selected model.
Documentation and Presentation
Introduction:
Briefly introduce the project, the dataset, and the objectives.
Data Importing and Preprocessing:
Explain how the data was imported and any preprocessing steps taken, including
handling missing values, outliers, and any transformations applied.
Exploratory Data Analysis (EDA):
Present summary statistics and initial visualizations (line plot, histogram, etc.) to
show the data’s distribution, trends, and seasonality.
Stationarity and Differencing:
Discuss stationarity tests and any differencing applied to the data to make it
stationary.
Moving Average and Exponential Smoothing:
Explain the moving average method, chosen window size, and smoothing parameters
for exponential smoothing. Present corresponding plots and compare them with the
original data.
ARIMA Modelling:
Describe the time series decomposition into trend, seasonality, and residuals.
Show autocorrelation and partial autocorrelation plots used to select ARIMA
parameters.
Present the fitted ARIMA model and diagnostic plots.
Model Evaluation and Comparison:
Present the evaluation metrics (MAE, MSE, RMSE, etc.) for each model.
Compare the performance of different models and explain which one was selected as
the best-performing model.
Forecasting and Predictions:
Show the forecasted values using the selected model.
Present any visualizations of future predictions and discuss their implications.
Conclusion:
Summarize the key findings, the best-performing model, and its effectiveness in
forecasting.
Discuss any limitations encountered and potential future work or improvements.
Deliverables :
1. PPT
2. Write up covering all the points.